本文在研究ＭＰＥＧ－２ＴＭ５建议的自适应量化策略的基础上，设计了一种新的自适应量化器。以块为基础分析宏块的局部视觉活动特性，并通过综合评价宏块中各块的视觉活动特性来最终决定自适应视觉量化因子。实验结果表明，本文所设计的自适应量化器能均匀分布图像编码主观失真，改善了图像质量，特别是减少了平坦区的块效应，降低了平坦区强边缘的失真。 Based on the research of the adaptive quantization strategy proposed by MPEG-2TM5, this paper designs a new adaptive quantizer. The local visual activities of macroblocks are analyzed on a block-by-block basis, and finally the adaptive visual quantization factors are determined by comprehensively evaluating the visual activity characteristics of each block in the macroblock. The experimental results show that the adaptive quantizer designed in this paper can distribute the subjective distortion of image coding evenly, improve the image quality, especially reduce the block effect in the flat area and reduce the distortion of the strong edge in the flat area.